Get the latest tech news
Diffusion Forcing: Next-Token Prediction Meets Full-Sequence Diffusion
Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion
In addition, one can rollout much longer videos with our method than the maximum sequence length it's trained on. We provide PNG visualizations below to reflect original quality of generated samples longer than training horizon. We provide PNG visualizations below to reflect original quality of generated samples longer than training horizon.
Or read this on Hacker News
